テクスチャにUV座標を設定できるようにし、スプライトシート読み込み時にハンドル数を節約できるように

[EbiseLutica]

fix #52

• Texture2Dに UvStart UvEnd` プロパティを追加
• OpenGLのテクスチャレンダラーが、 UvStart UvEnd プロパティを用いるよう変更
• OpenGLのテクスチャファクトリの LoadSpriteSheet が、本機能を用いて1つだけハンドルを生成し、複数のTexture2Dで使い回すように

[EbiseLutica]

NineSliceSpriteについても同様の処理を行うか？

[Copilot]

Pull Request Overview

Adds UV coordinate support to Texture2D, updates the OpenGL pipeline to use these UVs, and optimizes sprite sheet loading by reusing a single GPU texture handle.

• Introduce UvStart/UvEnd properties and overload constructors in Texture2D
• Update OpenGLTextureFactory.LoadSpriteSheet to generate one GL handle and slice UVs per sub-texture
• Modify GL renderer helper to stream vertex+UV data into a dynamic VBO per draw

Reviewed Changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
Promete/Graphics/Texture2D.cs	Added UV coordinate properties and new constructors
Promete/Windowing/GLDesktop/OpenGLTextureFactory.cs	New sprite sheet loader uses UV slicing and single GL handle
Promete/Nodes/Renderer/GL/Helper/GLTextureRendererHelper.cs	Switched VBO to dynamic draw and inject per-sprite UVs
Promete/Promete.csproj	Bumped package version
Promete/Graphics/FrameBuffer.cs	Removed unused using directives
Promete.Example/examples/graphics/sprite_sheet.cs	New demo showing sprite sheet loading and usage

Comments suppressed due to low confidence (4)

Promete.Example/examples/graphics/sprite_sheet.cs:9

• [nitpick] Class names should follow PascalCase. Rename sprite_sheet to SpriteSheetDemo or similar for consistency.

public class sprite_sheet : Scene

Promete/Graphics/Texture2D.cs:24

• [nitpick] Per .NET naming conventions, acronyms longer than one letter should be uppercase. Rename UvStart/UvEnd to UVStart/UVEnd.

    public Vector UvStart { get; }

Promete/Windowing/GLDesktop/OpenGLTextureFactory.cs:80

• The new sprite sheet loader logic isn’t covered by any existing tests. Consider adding unit or integration tests to verify correct UV slicing and exception cases.

    private Texture2D[] LoadSpriteSheet(Image bmp, int horizontalCount, int verticalCount, VectorInt size)

Promete/Promete.csproj:13

• [nitpick] Verify that the new pre-release suffix (-exp.2) aligns with your project’s versioning strategy and CI/CD processes.

        <Version>1.0.1-exp.2</Version>

[Copilot]

The call to LoadFromImageSharpImage(bmp) allocates a Texture2D that never gets disposed, leaking a GPU handle. Capture the returned Texture2D, retrieve its Handle, then call Dispose() on it explicitly.

Suggested change

	var handle = LoadFromImageSharpImage(bmp).Handle;
	var texture = LoadFromImageSharpImage(bmp);
	var handle = texture.Handle;
	texture.Dispose();

[Copilot]

Using ArgumentException with a null message and parameter name "horizontalCount" is confusing here. Prefer ArgumentOutOfRangeException with a clear message or use the correct parameter name (e.g., size or computed index).

Suggested change

	if (px + size.X > width) throw new ArgumentException(null, nameof(horizontalCount));
	if (py + size.Y > height) throw new ArgumentException(null, nameof(verticalCount));
	if (px + size.X > width) throw new ArgumentOutOfRangeException(nameof(size), $"Computed X index ({px + size.X}) exceeds image width ({width}).");
	if (py + size.Y > height) throw new ArgumentOutOfRangeException(nameof(size), $"Computed Y index ({py + size.Y}) exceeds image height ({height}).");

[Copilot]

Manual Dispose at the end won’t run if an exception is thrown earlier. Wrap the Image in a using or try/finally to guarantee disposal on error.

Suggested change

	var width = (float)bmp.Width;
	var height = (float)bmp.Height;
	var handle = LoadFromImageSharpImage(bmp).Handle;
	var textures = new Texture2D[verticalCount * horizontalCount];
	for (var y = 0; y < verticalCount; y++)
	{
	for (var x = 0; x < horizontalCount; x++)
	{
	var px = x * size.X;
	var py = y * size.Y;
	if (px + size.X > width) throw new ArgumentException(null, nameof(horizontalCount));
	if (py + size.Y > height) throw new ArgumentException(null, nameof(verticalCount));
	var uvStart = new Vector(px / width, py / height);
	var uvEnd = new Vector((px + size.X) / width, (py + size.Y) / height);
	textures[y * horizontalCount + x] = new Texture2D(handle, size, DisposeTexture, uvStart, uvEnd);
	}
	}
	bmp.Dispose();
	return textures;
	using (bmp)
	{
	var width = (float)bmp.Width;
	var height = (float)bmp.Height;
	var handle = LoadFromImageSharpImage(bmp).Handle;
	var textures = new Texture2D[verticalCount * horizontalCount];
	for (var y = 0; y < verticalCount; y++)
	{
	for (var x = 0; x < horizontalCount; x++)
	{
	var px = x * size.X;
	var py = y * size.Y;
	if (px + size.X > width) throw new ArgumentException(null, nameof(horizontalCount));
	if (py + size.Y > height) throw new ArgumentException(null, nameof(verticalCount));
	var uvStart = new Vector(px / width, py / height);
	var uvEnd = new Vector((px + size.X) / width, (py + size.Y) / height);
	textures[y * horizontalCount + x] = new Texture2D(handle, size, DisposeTexture, uvStart, uvEnd);
	}
	}
	return textures;
	}

[Copilot]

[nitpick] Uploading vertex data per sprite draw can become a bottleneck when rendering many sprites. Consider batching multiple sprites or using instanced rendering to reduce buffer updates.

Suggested change

	// VBOへ頂点データを書き込み
	gl.BindBuffer(BufferTargetARB.ArrayBuffer, _vbo);
	gl.BufferSubData<float>(BufferTargetARB.ArrayBuffer, 0, vertices);
	// バッチ用の頂点データを蓄積
	_batchedVertices.AddRange(vertices.ToArray());
	// バッチが一定サイズに達した場合、VBOへ頂点データを書き込み
	if (_batchedVertices.Count >= _batchSize)
	{
	gl.BindBuffer(BufferTargetARB.ArrayBuffer, _vbo);
	gl.BufferSubData<float>(BufferTargetARB.ArrayBuffer, 0, _batchedVertices.ToArray());
	_batchedVertices.Clear();
	}

[Copilot]

Only the first sub-texture is disposed, leaking the rest. Either call Dispose on all textures or implement a proper cleanup for the shared handle.

Suggested change

	_textures[0].Dispose();
	foreach (var texture in _textures)
	{
	texture.Dispose();
	}

[EbiseLutica]

_texturesのすべてのテクスチャのハンドルIDが共通しているので、最初のものだけを削除する対応でokなはず

[EbiseLutica]

macOS上で動かしたときに、タイルマップの処理が異常に遅くなってしまうようになった

実装に問題がありそうなので一旦マージは見送る（Draft扱いとする）